Client/Server with Distributed Annex?

Client/Server with Distributed Annex?

[Dale Stanbrough]

After looking at ACT's Glade, and the distributed systems annex, I became
unsure of whether the DS annex would allow for client/server programs where
clients come and go dynamically, but the server lives on. I got the feeling
that it was all a bit too static for this to be the case.

Any comments?


Dale

Re: Client/Server with Distributed Annex?

[Laurent Pautet]

>>>>> "Dale" == Dale Stanbrough <dale@goanna.cs.rmit.edu.au> writes:

Dale> After looking at ACT's Glade, and the distributed systems annex,
Dale> I became unsure of whether the DS annex would allow for
Dale> client/server programs where clients come and go dynamically,
Dale> but the server lives on. I got the feeling that it was all a bit
Dale> too static for this to be the case.

This feature has already been used by some companies to implement some
kind of replication. Yes, the DS annex allows to restart clients even
when some of them have died. It can use a mechanism based on RACWs,
but other approaches with plain RPCs are also possible if the server
doesn't need to notify or to keep tracks of its clients.

Have a look at DSA/Examples/Eratho/dynamic. Partitions 2, 3 and 4 are
in some way the clients you are talking about.

We can add such example in GLADE release if people are interested in
this kind of example.

-- 
-- Laurent